Infrequently Slow Server when using Talk

Nextcloud version: 24.0.6
Operating system and version: Ubuntu 20.04
Apache or nginx version (eg, Apache 2.4.25): Apache 2.4
PHP version (eg, 7.4): 7.4.3

The issue you are facing:
Server will run fine for days, maybe even a week. Then it will lock up all web site views to an extremely slow response time, if any. I can disable the Talk app then everything start running fine again. I can reenable Talk and it will also work fine for several days. I have watch logs on my server and htop as well with no obvious activity during these slow down times. Also I set up a TURN server and it shows very little resource use even during the times of slow down. Not sure if there is something else to check.

Is this the first time you’ve seen this error? (Y/N): N

Steps to replicate it:

Just seems that general use of Talk, even just messaging back and forth, can trigger the response issue.


For anyone tracking… This may be an issue of usage and settings regarding php-fpm. That is the area I’m targeting now.

Maybe check out Server system requirements - Nextcloud Talk API documentation ?!

These symptoms sound similar to an issue I was (hopefully “was”) facing where everything seemed slow, and during the slowness I noticed many slow queries against the db doing INSERT and DELETE against the oc_mounts table.

Have you noticed anything going on in the db while everything feels slow? Maybe try running SHOW PROCESSLIST over and over (assuming you are using MariaDB or MySQL).

Here’s the issue I mentioned: